At one time, I did run BackupPC with its pool over NFS. Though it was BackupPC running in a Linux VM on a Solaris server....with a zfs dataset from its mirrored rpool.
I didn't have any issues with uid/gid....since it was a typical NFSv3 share (root squash in effect), everything in the pool is normally owner/group of backuppc. .. so the uid/gid stayed the same, they jump mapped to a different user/group on the Solaris side. Don't know what Drobo does....a guess is that it might be doing a forced anonuid/anongid, so that you don't have to have consistent uid/gid mappings across your network. Someday I should make my uid/gid's consistent across my Linux/FreeBSD systems at home....perhaps when I complete getting them all under the control of CFEngine. I later got BackupPC running directly on Solaris. And, now its running on FreeBSD with its own zpool. I had done VM + NFS, because "apt-get install backuppc" was easier than customizing BackupPC to work on Solaris, but I've been customizing it here and there. Since I use varying incremental and full periods, depending on the data. It is hard to tell if the next backup of a given 'host' is imminent and whether its going to do a full or incremental when its time comes. So, this morning I tweaked the host summary page, tonight I plan to convert it into a CFEngine edit_line bundle :) Someday I'll get my work machine under the control of my CFEngine at home.... I've discovered that my failsafe.cf doesn't work from a clean slate....of a non-bootstrapped host (kind of hard to get the bootstrap to bring up a secure tunnel to discover my policy server at home....though failsafe has problems in this area as well...plus I wonder if all my remotes appear as the IP of my router is going to be a problem.) At work its 2x1TB mirrored. At home its 6x2TB's in a raidz2 (though the most common outage results in me losing 3 drives....since they are split between 2 - 5 bay enclosures.....have thought about whether a double conversion UPS is a consideration. When there's a flicker....sometimes it triggers a bus reset, and if power transfers back before the reset recovers....) Before FreeBSD, it was Ubuntu and they were managed by mdadm as a RAID 10.....but I did an upgrade from 8.04LTS to 10.04LTS....which had a bug that made some of my mdadm arrays vanish. The patch for the bug came out the following week. Before I built this big array, I had only been doing RAID1's...there were 5....I remember that the 6 disk array was md5, and known as vg4 (lv was backuppc.) I had played with 4x2TB in RAID5 briefly before jumping to the 6x2TB RAID10....which worked fine until I started having bitrot. The other reason I went big with FreeBSD at home, was that SLA customer at work had gone with such a system....so decided it was time I got back up to speed with it, etc. Though maybe I wasn't that behind....Netbackup client is for 6. Had talked about doing ZFS at home someday for years though....Windows 7 eating itself created an opening....(it auto applied a bunch of patches and then wouldn't boot, and the a chkdsk made everything vanish....always bugged me that Intel's RAID implied that when its 'initializing' my array is what most of us call resilvering. And, it does this only after after a BSOD. Their system is scary....FreeBSD 9 booting from a raidz of 6 2TB drives. But, its one of two production FreeBSD servers doing this (the second has a 100GB SSD for L2ARC....they had gotten thinking ZIL, but everything strong advises against not using mirrored SSDs for ZIL (especially a consumer MLC drive....maybe if they had gone with something like the 500GB Enterprise SLC drives in our 7420 :) Plus they don't need that much SSD for ZIL. I have 4G ZIL's at home, but the most I've seen in utilization is ~100MB. Guessing I'm no where near the interface max...6Gbps. :( Probably need to decide soon what I'm going to do with my remaining 10.04LTS server (there were two, but one died over Thanksgiving...so the last things on it got moved to a 12.04LTS box next to it....the rest had already been moved to a pair of FreeBSD servers....which I have discovered that I didn't leave space to do HAST as I original intended (trying to find details on how people are using zvols for HAST/CARP.) Though it might not matter with the 10.04LTS server....if more of its disks die off. Already lost over 1.5TB of data (well, its all in my BackupPC pool still) when a pair of ST2000DM000's decided they had sufficiently exceeded their 1 year warranty die close together. Didn't seem to matter that I had gotten them about 3 months apart from different sellers. I originally got one to replace a failed disk, which it reminded me the problems misaligned accesses on advanced format drives. At first I tweaked the partition which seemed to help....but eventually I turned it into two degraded arrays, copied the data across and moved the old disk into the new array.... Though wonder if using GPT would've caused a problem with upgrading ubuntu? Guess that's one less thing to worry about.... ;) The other alternative would've been if I got the system configuration documented in CFEngine....but I keep finding new things about CFEngine and the systems I do have under its control. On 05/09/14 09:19, Morgan Blackthorne wrote: > I was wondering if anyone had done this, and stored the pool over a redundant > array of drives (like in my case, on 2x4TB drives in a Drobo FS) via an NFS > mount that root can write to. I've noticed that if I copy something as a user, > it strips off the UID/GID with a warning, but I'm not sure if that's something > that would actually impact the way that BackupPC operates. Given that it does > deduping, I think it already has an index of file metadata and a reference to > where to find it inside the pool. > > But I also don't want to screw up a working system, either, so wanted to see > if anyone might know of some pitfalls ahead in this prospect. Worst case, I > could just move some disks around now that I've expanded the Drobo, and mirror > two 1TB drives just for the backup purposes with mdadm. > > Thanks for any advice. > > -- > ~*~ StormeRider ~*~ > > "Every world needs its heroes [...] They inspire us to be better than we are. > And they protect from the darkness that's just around the corner." > > (from Smallville Season 6x1: "Zod") > > On why I hate the phrase "that's so lame"... http://bit.ly/Ps3uSS > > > _______________________________________________ > Discuss mailing list > [email protected] > https://lists.lopsa.org/cgi-bin/mailman/listinfo/discuss > This list provided by the League of Professional System Administrators > http://lopsa.org/ > -- Who: Lawrence K. Chen, P.Eng. - W0LKC - Sr. Unix Systems Administrator For: Enterprise Server Technologies (EST) -- & SafeZone Ally _______________________________________________ Discuss mailing list [email protected] https://lists.lopsa.org/cgi-bin/mailman/listinfo/discuss This list provided by the League of Professional System Administrators http://lopsa.org/
